Designing for a modern and resilient analytics stack is not only about building for an intelligent data architecture via cloud native technologies or ensuring that an advanced analytics platform has a comprehensive set of data connectors with interactive data visualization tools. Nor is it about encompassing the latest augmented artifical intelligence capabilities. Rather it is striving towards a holistic framework that brings people, processes, data nad technology together for the best business outcomes.
First and foremost, a strong foundation in data calls for applying a business oriented mindset to identifying the capabilities a business is looking for. It is necessary to define business triggers and review high impact use cases while ensuring awareness and alignment with business and IT stakeholders to achieve the intended results. Once the motivations, objectives and expected benefits are assessed, the next stage calls for a full assessment on the current analytics platform or digital data estate and its components both at a high level of understanding and a detailed level of evaluation. It is necessary to emphasize that a culture change around the perception of data as a source for reporting needs to change.
Data governance often focuses on data quality, data security and access and data lifecycle management. With the release of the General Data Protection Regulation in view of rising concerns on the lack of data quality and privacy breaches arising from poorly governed data production and exchange , Establishing a governance model is often conditional upon the business maturity and sophistication of an organization. Highly regulated industries will often have the most robust and comprehensive data governance model as data regulations require additional controls and reporting. Hence the opportunity for metadata automation and data masking for privacy controls whether in exploration mode or broader customer setting will need to be strongly enforced.
From a broad perspective data governance can be viewed from a series of stages. The most basic level starts with information profiling which involves data quality and lineage and proveanance. It might also involve data inventory and analysis. Next, is about privacy and security. Examples include information sensitivity classification, privacy management and role-based access controls. Compliance would include the regulatory aspect, corporate IT policices and procedures. The last stage is about creating awareness through employee training and engagement and having visibility.
Data is processed in siloed and not shared between departments. This means there is a lack of accountability and responsibility for data related issues and a subsequent lack of visibility to mitigate the potential risks
When data becomes obsolete, its value to return actionable insights decreases. Data goverannce aims to improve the following attributes :
Availability Usability Integrity Security
Model governance sets the context around how organizations implement and operate machine learning in production while balancing efficiencies and achieving more with their AI investments. With improved governance comes improved data quality, stronger security and better decision making